pith. sign in

arxiv: 2605.19366 · v1 · pith:PCVRGVLTnew · submitted 2026-05-19 · 💻 cs.LG

Accurate, Efficient, and Explainable Deep Learning Approaches for Environmental Science Problems

Jimeng Shi This is my paper

Pith reviewed 2026-05-20 07:56 UTC · model grok-4.3

classification 💻 cs.LG
keywords deep learningenvironmental scienceflood predictionweather forecastingquestion answeringdiffusion modelsretrieval augmented generationexplainable AI
0
0 comments X

The pith

Deep learning models deliver accurate, efficient, and explainable solutions for flood prediction, weather forecasting, and environmental question answering.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper develops three deep learning methods to handle large-scale environmental data challenges that traditional physics models struggle with due to computational cost or lack of uncertainty estimates. For coastal flood systems with extreme rainfall and sea level changes, FIDLAr forecasts and manages water levels while outperforming baselines and providing interpretable outputs. CoDiCast adapts diffusion models to produce probabilistic global weather forecasts efficiently and with explicit uncertainty. Hypercube-RAG uses a structured retrieval framework so language models answer domain questions accurately without hallucinations while remaining efficient and explainable. These approaches matter because they support faster, more transparent decisions for protecting ecosystems and managing natural resources.

Core claim

The dissertation claims that specialized deep learning architectures can simultaneously achieve accuracy, efficiency, and explainability across environmental tasks where physics-based methods fall short: FIDLAr outperforms baselines in accuracy and efficiency for water level forecasting and management in a South Florida coastal system while providing interpretable outputs; CoDiCast, a conditional diffusion model, achieves accurate and efficient probabilistic weather forecasts with uncertainty quantification; Hypercube-RAG, built on a structured text cube, exhibits accuracy, efficiency, and explainability at once for scientific question answering in environmental science.

What carries the argument

The key machinery consists of three tailored deep learning constructions: a forecast-informed model for water level control, a conditional diffusion model for probabilistic prediction, and a hypercube-structured retrieval-augmented generation system that retrieves domain knowledge to ground answers.

If this is right

  • FIDLAr enables real-time water level management in flood-prone coastal areas where physics simulations are too slow.
  • CoDiCast supplies probabilistic weather forecasts that include uncertainty measures for improved planning.
  • Hypercube-RAG reduces hallucinations in answers to environmental science questions by grounding outputs in retrieved domain knowledge.
  • The three approaches together reduce dependence on computationally heavy traditional models for operational environmental decisions.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • If the models generalize beyond the tested regions, they could form the basis for operational environmental intelligence systems worldwide.
  • Combining these data-driven methods with select physical constraints could produce hybrid models that are both faster and more robust.
  • The structured retrieval idea in Hypercube-RAG might transfer to question answering in other data-rich scientific fields facing similar knowledge gaps.

Load-bearing premise

The proposed deep learning architectures can capture the governing dynamics of complex environmental systems like extreme rainfall, sea level changes, and global atmospheric patterns well enough to outperform physics-based models without added physical constraints.

What would settle it

Running FIDLAr or CoDiCast on a new extreme rainfall or weather event outside the training distribution and finding prediction errors larger than those from established physics models would falsify the performance claims.

Figures

Figures reproduced from arXiv: 2605.19366 by Jimeng Shi.

Figure 2.1
Figure 2.1. Figure 2.1: Schematic representation of the MLP architecture with [PITH_FULL_IMAGE:figures/full_fig_p032_2_1.png] view at source ↗
Figure 2.2
Figure 2.2. Figure 2.2: Schematic representation of the RNN architecture with its unfolded represen [PITH_FULL_IMAGE:figures/full_fig_p033_2_2.png] view at source ↗
Figure 2.3
Figure 2.3. Figure 2.3: Schematic representation of the convoluntion operation in CNNs [CTCO19]. [PITH_FULL_IMAGE:figures/full_fig_p034_2_3.png] view at source ↗
Figure 2.4
Figure 2.4. Figure 2.4: Schematic representation of the Attention mechanism [VSP [PITH_FULL_IMAGE:figures/full_fig_p036_2_4.png] view at source ↗
Figure 2.5
Figure 2.5. Figure 2.5: Schematic representation of the DDPM [HJA20]. [PITH_FULL_IMAGE:figures/full_fig_p037_2_5.png] view at source ↗
Figure 2.6
Figure 2.6. Figure 2.6: Schematic representation of the output process of LLMs. [PITH_FULL_IMAGE:figures/full_fig_p038_2_6.png] view at source ↗
Figure 2.7
Figure 2.7. Figure 2.7: Taxonomy of deep learning models for weather prediction across training [PITH_FULL_IMAGE:figures/full_fig_p043_2_7.png] view at source ↗
Figure 2
Figure 2. Figure 2: , including training paradigms, model architectures, and scopes. [PITH_FULL_IMAGE:figures/full_fig_p044_2.png] view at source ↗
Figure 3.1
Figure 3.1. Figure 3.1: Forecast-Informed Deep Learning Architecture (FIDLA [PITH_FULL_IMAGE:figures/full_fig_p051_3_1.png] view at source ↗
Figure 3.2
Figure 3.2. Figure 3.2: Flood Evaluator. The parts shaded green are used as inputs (i.e., historical data and covariates predicted from the near future) and orange (control schedule for the gates and pumps) Water levels (blue) are the outputs. 3.3.4 Flood Manager Flood Manager is to produce control schedules for hydraulic structures (i.e., gates and pumps), taking as inputs reliably predictable future information (rain, tide) a… view at source ↗
Figure 3.3
Figure 3.3. Figure 3.3: Flood Manager. The parts shaded green (historical data) are the inputs, and the parts shaded orange are the outputs. The water levels shaded blue are not predicted. The resulting output of water levels can be used to compute the loss in Eq. (3.6), representing evaluation scores for generated control schedules. Gradient descent [Rud16] can be back-propagated as the feedback to update the parameters of the… view at source ↗
Figure 3.4
Figure 3.4. Figure 3.4: The two red bars represent a threshold of flooding and a threshold of water [PITH_FULL_IMAGE:figures/full_fig_p055_3_4.png] view at source ↗
Figure 3.5
Figure 3.5. Figure 3.5: Graph Transformer Network for Flood Evaluator. Input variables in￾clude Rainfall, Pump, Gate, Tide, and Water levels as shown in [PITH_FULL_IMAGE:figures/full_fig_p056_3_5.png] view at source ↗
Figure 3.6
Figure 3.6. Figure 3.6: Schematic diagram of study domain. There are three water stations with [PITH_FULL_IMAGE:figures/full_fig_p057_3_6.png] view at source ↗
Figure 3.7
Figure 3.7. Figure 3.7: Visualization of water levels with various methods for flood mitigation. [PITH_FULL_IMAGE:figures/full_fig_p061_3_7.png] view at source ↗
Figure 3.8
Figure 3.8. Figure 3.8: Visualization for flood mitigation at all locations. 3.6.3 Ablation Study As introduced in [PITH_FULL_IMAGE:figures/full_fig_p061_3_8.png] view at source ↗
Figure 3.9
Figure 3.9. Figure 3.9: Importance scores of tide input. x and y axes are the tide and control schedule of the gate over time. 3.7 Discussion 3.7.1 Model Explainability The explainability feature, which is shown with an example in [PITH_FULL_IMAGE:figures/full_fig_p064_3_9.png] view at source ↗
Figure 4.1
Figure 4.1. Figure 4.1: Deterministic vs Probabilistic Models. 4.3.2 Denoising Diffusion Probabilistic Models A denoising diffusion probabilistic model (DDPM) [HJA20] generates target samples by learning a distribution pθ(x0) that approximates the target distribution q(x0). DDPM comprises a forward diffusion process and a reverse denoising process. The forward process transforms an input x0 with a data distribution of q(x0) to … view at source ↗
Figure 4
Figure 4. Figure 4: illustrates the overall framework of our proposed approach, C [PITH_FULL_IMAGE:figures/full_fig_p075_4.png] view at source ↗
Figure 4.2
Figure 4.2. Figure 4.2: Framework of our conditional diffusion model for global weather forecast [PITH_FULL_IMAGE:figures/full_fig_p076_4_2.png] view at source ↗
Figure 4.3
Figure 4.3. Figure 4.3: Autoencoder structure. 4.4.5 Attention-based Denoiser Network Our denoiser network consists of two blocks: cross-attention and U-net (as shown in [PITH_FULL_IMAGE:figures/full_fig_p077_4_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: ). Cross-attention mechanism [HMT [PITH_FULL_IMAGE:figures/full_fig_p077_4.png] view at source ↗
Figure 4.4
Figure 4.4. Figure 4.4: Attention-based denoiser structure. U-Net [RFB15] is utilized to recover the data by removing the noise added at each dif￾fusion step. The skip connection technique in U-Net concatenates feature maps from the encoder to the corresponding decoder layers, allowing the network to retain fine-grained information that might be lost during downsampling. The detailed U-Net architecture is presented in [PITH_FU… view at source ↗
Figure 4.5
Figure 4.5. Figure 4.5: Architecture of the U-Net model. loss function: Lcond(θ) = EX0,ϵ,n [PITH_FULL_IMAGE:figures/full_fig_p079_4_5.png] view at source ↗
Figure 4.4
Figure 4.4. Figure 4.4: Algorithm 2 Pseudocode for Training Process 1: Input: Number of diffusion steps N, pre-trained encoder F 2: Output: Trained denoising function ϵ(·) 3: repeat 4: X t+1 0 ∼ q(X t+1 0 ) 5: n ∼ Uniform(1, 2, . . . , N) 6: ϵ ∼ N (0, I) 7: Get the past observations Xt−1 , Xt 8: Get embedding Z˜t−1:t = F(Xt−1 , Xt ) 9: Take gradient descent step on: ∇θ [PITH_FULL_IMAGE:figures/full_fig_p079_4_4.png] view at source ↗
Figure 4.6
Figure 4.6. Figure 4.6: Model Forecasts with confidence intervals. [PITH_FULL_IMAGE:figures/full_fig_p087_4_6.png] view at source ↗
Figure 4.7
Figure 4.7. Figure 4.7: Visualizations of true and predicted values of all five variables at 24 hours [PITH_FULL_IMAGE:figures/full_fig_p089_4_7.png] view at source ↗
Figure 4.8
Figure 4.8. Figure 4.8: Visualization of true and predicted values across five meteorological variables [PITH_FULL_IMAGE:figures/full_fig_p089_4_8.png] view at source ↗
Figure 4.9
Figure 4.9. Figure 4.9: Visualizations of true and predicted values of all five variables at 144 hours [PITH_FULL_IMAGE:figures/full_fig_p090_4_9.png] view at source ↗
Figure 4.10
Figure 4.10. Figure 4.10: Ablation study. 4.6.4 Parameter Study [PITH_FULL_IMAGE:figures/full_fig_p090_4_10.png] view at source ↗
Figure 4.11
Figure 4.11. Figure 4.11: Effect of linear and quadratic variance scheduling methods. [PITH_FULL_IMAGE:figures/full_fig_p091_4_11.png] view at source ↗
Figure 5.1
Figure 5.1. Figure 5.1: Hypecube- vs semantic RAG: A case study atively improved, the hurdle of missing specific themes remains, causing semantically similar but off-topic retrievals [KRI21, KAJ+24] [PITH_FULL_IMAGE:figures/full_fig_p094_5_1.png] view at source ↗
Figure 5.2
Figure 5.2. Figure 5.2: Graph vs. Hypercube. The methods discussed above tend to perform poorly in at least one of the aspects among accuracy, efficiency, and explainability. This motivates us to develop a RAG system that overcomes this weakness. In this pursuit, we identified the text cube as a promising technique [TZC+18, WJH+23]. A text cube is an inherently explainable mul￾tidimensional structure that allocates documents in… view at source ↗
Figure 5
Figure 5. Figure 5: illustrates the [PITH_FULL_IMAGE:figures/full_fig_p098_5.png] view at source ↗
Figure 5.4
Figure 5.4. Figure 5.4: Hypercube construction on a corpus. We present only three dimensions for [PITH_FULL_IMAGE:figures/full_fig_p099_5_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: ), then retrieval inside of hypercube is performed by matching these decom [PITH_FULL_IMAGE:figures/full_fig_p101_5.png] view at source ↗
Figure 5.5
Figure 5.5. Figure 5.5: Prompt template for LLM-as-a-judge. 5.5 Results In this section, we show and analyze the experimental results for accuracy, efficiency, and explainability of our Hypercube-RAG and other baseline methods. 5.5.1 Accuracy We compare our Hypercube-RAG with semantic RAG methods, graph-based RAG methods, and LLMs without retrieval [PITH_FULL_IMAGE:figures/full_fig_p106_5_5.png] view at source ↗
Figure 5.6
Figure 5.6. Figure 5.6: Performance comparison with various LLMs (only the range between 50% and 90% shown). higher retrieval costs. This is particularly pronounced for graph-based methods due to the computational burden of search paths in large-scale graphs, posing scalability challenges. Notably, our Hypercube-RAG substantially reduces the retrieval time by one to two orders of magnitude compared to both semantic and graph-ba… view at source ↗
Figure 5.7
Figure 5.7. Figure 5.7: Ablation study on constituents of Hypercube-RAG. We also conduct an ablation study on the hypercube dimensions: Location, Event, Date, Organization, Person, and Theme. To study their effectiveness, we con￾duct an ablation study by removing each of the dimensions, which are represented as No-Location, No-Event, No-Date, No-Organization, No-Person, and No-Theme, respectively [PITH_FULL_IMAGE:figures/full_… view at source ↗
Figure 5.8
Figure 5.8. Figure 5.8: Performance vs similarity threshold. 5.5.6 Case Study An additional case study in [PITH_FULL_IMAGE:figures/full_fig_p114_5_8.png] view at source ↗
Figure 5.9
Figure 5.9. Figure 5.9: Comparison of three RAG methods on the same query. [PITH_FULL_IMAGE:figures/full_fig_p115_5_9.png] view at source ↗
Figure 5.10
Figure 5.10. Figure 5.10: Access one cube cell in one hypercube. ✓ represents the touched cube cells. Case 2: long query with multiple topics. In cases where a query is very diverse, they may need to access multiple cube cells such that the query information can be covered 100 [PITH_FULL_IMAGE:figures/full_fig_p116_5_10.png] view at source ↗
Figure 5.11
Figure 5.11. Figure 5.11: Access multiple cube cells in one hypercube. [PITH_FULL_IMAGE:figures/full_fig_p117_5_11.png] view at source ↗
Figure 5.12
Figure 5.12. Figure 5.12: Access multiple cube cells in multiple hypercubes. [PITH_FULL_IMAGE:figures/full_fig_p118_5_12.png] view at source ↗
read the original abstract

Environmental science plays a pivotal role in safeguarding ecosystems, a domain driven by large-scale, heterogeneous data. In the big data era, artificial intelligence (AI) has emerged as a transformative tool for learning patterns and supporting decision-making. This dissertation develops AI-based approaches tailored to complex environmental science problems to achieve Environmental Intelligence, studying three specific challenges. First, we focus on flood prediction and management in coastal river systems. Conventional physics-based models are computationally intensive, limiting real-time application. To overcome this, we propose a deep learning (DL)-based model, WaLeF, for water level forecasting, and a forecast-informed DL model, FIDLAr, to manage water levels. Evaluated in a flood-prone coastal system in South Florida characterized by extreme rainfall and sea level fluctuations, FIDLAr outperforms baselines in accuracy and efficiency while providing interpretable outputs. Second, we target global weather prediction, which is challenged by massive data scale. Traditional physics methods are deterministic and computationally heavy. We propose CoDiCast, a conditional diffusion model tailored for probabilistic weather forecasting. Adapted from generative AI for predictive tasks, experiments show CoDiCast achieves accurate, efficient forecasts with explicit uncertainty quantification. Lastly, we address scientific question-answering in environmental science. When answering in-domain questions, large language models (LLMs) often suffer from hallucinations due to out-of-date or limited knowledge. While retrieval-augmented generation (RAG) retrieves domain-specific knowledge, existing methods trade off accuracy, efficiency, or explainability. We propose Hypercube-RAG, built on a structured text cube framework, which successfully exhibits all three properties simultaneously.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

3 major / 2 minor

Summary. The manuscript develops three deep learning approaches for environmental science challenges: WaLeF and FIDLAr for real-time water level forecasting and flood management in a South Florida coastal system, CoDiCast as a conditional diffusion model for probabilistic global weather forecasting, and Hypercube-RAG as a structured text-cube retrieval-augmented generation framework for domain-specific scientific question answering. The central claims are that these methods deliver improved accuracy and efficiency over physics-based baselines while adding interpretability or uncertainty quantification, without relying on conventional physical constraints.

Significance. If the empirical results hold under rigorous scrutiny, the work could support more scalable Environmental Intelligence tools for real-time decision support in flood management and weather prediction, with Hypercube-RAG addressing hallucination issues in LLMs for scientific QA. The explicit uncertainty in CoDiCast and explainability focus in Hypercube-RAG are positive differentiators from standard DL applications. However, the significance depends on whether data-driven models can reliably capture heterogeneous environmental dynamics without introducing non-physical artifacts.

major comments (3)
  1. [Abstract and §3] Abstract and §3 (FIDLAr): The claim that FIDLAr outperforms physics-based models in accuracy and efficiency for extreme rainfall and sea-level scenarios is presented without reported checks for physical consistency (e.g., mass conservation in water-level predictions or adherence to known hydrodynamic bounds during out-of-distribution events). Standard accuracy metrics alone do not rule out non-physical artifacts, which is load-bearing for the assertion that the architecture reliably learns governing dynamics from data alone.
  2. [§4] §4 (CoDiCast): The probabilistic forecasts are stated to provide explicit uncertainty quantification, yet the evaluation lacks calibration diagnostics (e.g., reliability diagrams or comparison against physics ensemble spreads) or long-horizon stability tests. This undermines the claim of accurate, efficient forecasts as a direct alternative to deterministic physics methods, especially given the skeptic concern about non-stationary drivers.
  3. [§5] §5 (Hypercube-RAG): While the structured text-cube framework is said to achieve accuracy, efficiency, and explainability simultaneously, the manuscript does not quantify the trade-off resolution (e.g., via ablation on retrieval latency vs. explanation fidelity or hallucination rate on held-out environmental queries). Without these metrics, the simultaneous achievement remains an unverified assertion.
minor comments (2)
  1. Dataset details (size, temporal resolution, train/test splits) for the South Florida flood case and global weather benchmarks should be tabulated for reproducibility.
  2. Notation for the conditional diffusion process in CoDiCast should be aligned with standard diffusion literature to avoid ambiguity in the forward/reverse steps.

Simulated Author's Rebuttal

3 responses · 0 unresolved

We thank the referee for their constructive comments on our manuscript arXiv:2605.19366. We address each of the major comments point by point below, indicating the revisions we plan to make.

read point-by-point responses

  1. Referee: [Abstract and §3] Abstract and §3 (FIDLAr): The claim that FIDLAr outperforms physics-based models in accuracy and efficiency for extreme rainfall and sea-level scenarios is presented without reported checks for physical consistency (e.g., mass conservation in water-level predictions or adherence to known hydrodynamic bounds during out-of-distribution events). Standard accuracy metrics alone do not rule out non-physical artifacts, which is load-bearing for the assertion that the architecture reliably learns governing dynamics from data alone.

    Authors: We recognize the referee's concern regarding the verification of physical consistency in our data-driven FIDLAr model. Although the model is designed to learn from data without explicit physical constraints, as stated in the manuscript, the superior performance on real observational data from the South Florida system provides indirect evidence of consistency. However, to directly address this, we will incorporate explicit physical consistency checks, such as mass conservation analysis and adherence to hydrodynamic bounds for extreme events, in the revised manuscript. revision: yes

  2. Referee: [§4] §4 (CoDiCast): The probabilistic forecasts are stated to provide explicit uncertainty quantification, yet the evaluation lacks calibration diagnostics (e.g., reliability diagrams or comparison against physics ensemble spreads) or long-horizon stability tests. This undermines the claim of accurate, efficient forecasts as a direct alternative to deterministic physics methods, especially given the skeptic concern about non-stationary drivers.

    Authors: We appreciate this observation on the evaluation of CoDiCast. To strengthen the claims of accurate probabilistic forecasting with explicit uncertainty quantification, we will add calibration diagnostics including reliability diagrams and comparisons to physics-based ensemble spreads. We will also include long-horizon stability tests to assess performance under non-stationary conditions, thereby providing a more rigorous validation against the concerns raised. revision: yes

  3. Referee: [§5] §5 (Hypercube-RAG): While the structured text-cube framework is said to achieve accuracy, efficiency, and explainability simultaneously, the manuscript does not quantify the trade-off resolution (e.g., via ablation on retrieval latency vs. explanation fidelity or hallucination rate on held-out environmental queries). Without these metrics, the simultaneous achievement remains an unverified assertion.

    Authors: We agree that quantifying the resolution of trade-offs is essential to support the claims for Hypercube-RAG. In the revised manuscript, we will include additional ablation studies that measure retrieval latency against explanation fidelity and hallucination rates on held-out environmental queries. This will provide concrete evidence of how the structured text-cube framework achieves the three properties simultaneously. revision: yes

Circularity Check

0 steps flagged

No significant circularity; claims rest on empirical evaluations of proposed models

full rationale

The paper proposes three new deep learning architectures (FIDLAr for flood management, CoDiCast for probabilistic weather forecasting, and Hypercube-RAG for scientific QA) and reports their performance via experiments on environmental datasets. No mathematical derivations, equations, or first-principles chains are presented in the abstract or described methods; outperformance claims are framed as direct empirical outcomes against baselines rather than reductions to fitted parameters or self-referential definitions. No load-bearing self-citations, uniqueness theorems, or ansatzes smuggled via prior work are invoked to justify core results. The work is self-contained against external benchmarks and does not exhibit any of the enumerated circularity patterns.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

Abstract-only review provides no equations, datasets, or modeling choices; no free parameters, axioms, or invented entities can be identified.

pith-pipeline@v0.9.0 · 5819 in / 1268 out tokens · 29021 ms · 2026-05-20T07:56:43.027773+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

What do these tags mean?
matches
The paper's claim is directly supported by a theorem in the formal canon.
supports
The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends
The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses
The paper appears to rely on the theorem as machinery.
contradicts
The paper's claim conflicts with a theorem or certificate in the canon.
unclear
Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.

Reference graph

Works this paper leans on

300 extracted references · 300 canonical work pages · 16 internal anchors

  1. [1]

    Flood risk assessment of the population in

    Ikram, Qiyamud Din and Jamalzi, Abdur Rashid and Hamidi, Abdur Rahim and Ullah, Irfan and Shahab, Muhmmad , journal=. Flood risk assessment of the population in. 2024 , publisher=

    work page 2024

  2. [2]

    Natural hazards-impacts, adjustments and resilience , pages=

    Flood disaster hazards; causes, impacts and management: a state-of-the-art review , author=. Natural hazards-impacts, adjustments and resilience , pages=. 2021 , publisher=

    work page 2021

  3. [3]

    Nature Reviews Earth & Environment , volume=

    Causes, impacts and patterns of disastrous river floods , author=. Nature Reviews Earth & Environment , volume=. 2021 , publisher=

    work page 2021

  4. [4]

    The state of the art in deep learning applications, challenges, and future prospects:

    Kumar, Vijendra and Azamathulla, Hazi Md and Sharma, Kul Vaibhav and Mehta, Darshan J and Maharaj, Kiran Tota , journal=. The state of the art in deep learning applications, challenges, and future prospects:. 2023 , publisher=

    work page 2023

  5. [5]

    Nature Water , pages=

    Strategic storm flood evacuation planning for large coastal cities enables more effective transfer of elderly populations , author=. Nature Water , pages=. 2024 , publisher=

    work page 2024

  6. [6]

    Proceedings of the 2023 SIAM International Conference on Data Mining (SDM) , pages=

    Physics-guided meta-learning method in baseflow prediction over large regions , author=. Proceedings of the 2023 SIAM International Conference on Data Mining (SDM) , pages=. 2023 , organization=

    work page 2023

  7. [7]

    International Journal of Disaster Risk Reduction , volume=

    Physics-based simulations of multiple natural hazards for risk-sensitive planning and decision making in expanding urban regions , author=. International Journal of Disaster Risk Reduction , volume=. 2023 , publisher=

    work page 2023

  8. [8]

    Machine learning for metal additive manufacturing:

    Guo, Shenghan and Agarwal, Mohit and Cooper, Clayton and Tian, Qi and Gao, Robert X and Grace, Weihong Guo and Guo, YB , journal=. Machine learning for metal additive manufacturing:. 2022 , publisher=

    work page 2022

  9. [9]

    Results in Engineering , volume=

    A review of physics-based machine learning in civil engineering , author=. Results in Engineering , volume=. 2022 , publisher=

    work page 2022

  10. [10]

    Differentiable modeling to unify machine learning and physical models and advance Geosciences.arXiv:2301.04027, May

    Differentiable modeling to unify machine learning and physical models and advance Geosciences , author=. arXiv preprint arXiv:2301.04027 , year=

    work page arXiv

  11. [11]

    CIRP Annals , volume=

    Digital twins for electro-physical, chemical, and photonic processes , author=. CIRP Annals , volume=. 2023 , publisher=

    work page 2023

  12. [12]

    Journal of hydrologic engineering , volume=

    An overview of flood concepts, challenges, and future directions , author=. Journal of hydrologic engineering , volume=. 2022 , publisher=

    work page 2022

  13. [13]

    Journal of Hydroinformatics , volume=

    Optimisation of a fuzzy logic-based local real-time control system for mitigation of sewer flooding using genetic algorithms , author=. Journal of Hydroinformatics , volume=. 2020 , publisher=

    work page 2020

  14. [14]

    Leon, Arturo S and Tang, Yun and Qin, Li and Chen, Duan , journal=. A. 2020 , publisher=

    work page 2020

  15. [15]

    Review on model predictive control:

    Schwenzer, Max and Ay, Muzaffer and Bergs, Thomas and Abel, Dirk , journal=. Review on model predictive control:. 2021 , publisher=

    work page 2021

  16. [16]

    IEEE Access , volume=

    Recent achievements in model predictive control techniques for industrial motor: A comprehensive state-of-the-art , author=. IEEE Access , volume=. 2021 , publisher=

    work page 2021

  17. [17]

    npj Computational Materials , volume=

    Recent advances and applications of deep learning methods in materials science , author=. npj Computational Materials , volume=. 2022 , publisher=

    work page 2022

  18. [18]

    Artificial intelligence:

    Xu, Yongjun and Liu, Xin and Cao, Xin and Huang, Changping and Liu, Enke and Qian, Sen and Liu, Xingchen and Wu, Yanjun and Dong, Fengliang and Qiu, Cheng-Wei and others , journal=. Artificial intelligence:. 2021 , publisher=

    work page 2021

  19. [19]

    International Journal of Machine Learning and Cybernetics , volume=

    Recent advances in deep learning , author=. International Journal of Machine Learning and Cybernetics , volume=. 2020 , publisher=

    work page 2020

  20. [20]

    International conference on engineering applications of neural networks , pages=

    A robust, quantization-aware training method for photonic neural networks , author=. International conference on engineering applications of neural networks , pages=. 2022 , organization=

    work page 2022

  21. [21]

    Proceedings of the IEEE , year=

    Training spiking neural networks using lessons from deep learning , author=. Proceedings of the IEEE , year=

    work page

  22. [22]

    Towards explainable deep neural networks (x

    Angelov, Plamen and Soares, Eduardo , journal=. Towards explainable deep neural networks (x. 2020 , publisher=

    work page 2020

  23. [23]

    2021 International Conference on Computational Intelligence and Computing Applications (ICCICA) , pages=

    Possibilities and apprehensions in the landscape of artificial intelligence in education , author=. 2021 International Conference on Computational Intelligence and Computing Applications (ICCICA) , pages=. 2021 , organization=

    work page 2021

  24. [24]

    Assessment of inundation risk in urban floods using

    Rangari, Vinay Ashok and Umamahesh, NV and Bhatt, CM , journal=. Assessment of inundation risk in urban floods using. 2019 , publisher=

    work page 2019

  25. [25]

    Case study of

    Dasallas, Lea and Kim, Yeonsu and An, Hyunuk , journal=. Case study of. 2019 , publisher=

    work page 2019

  26. [26]

    The use of

    Iosub, Marina and Minea, I and Hapciuc, Oana and Romanescu, GH , journal=. The use of

    work page

  27. [27]

    Application of latest

    Shaikh, Arbaaz Aziz and Pathan, Azazkhan Ibrahimkhan and Waikhom, Sahita Ibopishak and Agnihotri, Prasit Girish and Islam, Md Nazrul and Singh, Sudhir Kumar , journal=. Application of latest. 2023 , publisher=

    work page 2023

  28. [28]

    Simulation of one-dimensional dam-break flood routing based on

    Mo, Chongxun and Shen, Yue and Lei, Xingbi and Ban, Huazhen and Ruan, Yuli and Lai, Shufeng and Cen, Weiyan and Xing, Zhenxiang , journal=. Simulation of one-dimensional dam-break flood routing based on. 2023 , publisher=

    work page 2023

  29. [29]

    Modeling flash floods in ungauged mountain catchments of

    Ragettli, Silvan and Zhou, Jian and Wang, Haijing and Liu, Changjun and Guo, Liang , journal=. Modeling flash floods in ungauged mountain catchments of. 2017 , publisher=

    work page 2017

  30. [30]

    Water Resources Management , volume=

    Improvement of best first decision trees using bagging and dagging ensembles for flood probability mapping , author=. Water Resources Management , volume=. 2020 , publisher=

    work page 2020

  31. [31]

    Spatial prediction of flood susceptibility using random-forest and boosted-tree models in Seoul metropolitan city,

    Lee, Sunmin and Kim, Jeong-Cheol and Jung, Hyung-Sup and Lee, Moung Jin and Lee, Saro , journal=. Spatial prediction of flood susceptibility using random-forest and boosted-tree models in Seoul metropolitan city,. 2017 , publisher=

    work page 2017

  32. [32]

    Daily forecasting of dam water levels: comparing a support vector machine (

    Hipni, Afiq and El-shafie, Ahmed and Najah, Ali and Karim, Othman Abdul and Hussain, Aini and Mukhlisin, Muhammad , journal=. Daily forecasting of dam water levels: comparing a support vector machine (. 2013 , publisher=

    work page 2013

  33. [33]

    Predicting streamflow in

    Essam, Yusuf and Huang, Yuk Feng and Ng, Jing Lin and Birima, Ahmed H and Ahmed, Ali Najah and El-Shafie, Ahmed , journal=. Predicting streamflow in. 2022 , publisher=

    work page 2022

  34. [34]

    Exploring temporal dynamics of river discharge using univariate long short-term memory (

    Mehedi, Md Abdullah Al and Khosravi, Marzieh and Yazdan, Munshi Md Shafwat and Shabanian, Hanieh , journal=. Exploring temporal dynamics of river discharge using univariate long short-term memory (. 2022 , publisher=

    work page 2022

  35. [35]

    Prediction of flow based on a

    Li, Peifeng and Zhang, Jin and Krebs, Peter , journal=. Prediction of flow based on a. 2022 , publisher=

    work page 2022

  36. [36]

    2016 , publisher=

    Smarter stormwater systems , author=. 2016 , publisher=

    work page 2016

  37. [37]

    Journal of Hydrology , volume=

    Exploring real-time control of stormwater systems for mitigating flood risk due to sea level rise , author=. Journal of Hydrology , volume=. 2020 , publisher=

    work page 2020

  38. [38]

    Journal of Hydrology , volume=

    An integrated optimization and rule-based approach for predictive real time control of urban stormwater management systems , author=. Journal of Hydrology , volume=. 2019 , publisher=

    work page 2019

  39. [39]

    2018 IEEE 43rd Conference on Local Computer Networks Workshops (LCN Workshops) , pages=

    A belief rule based flood risk assessment expert system using real time sensor data streaming , author=. 2018 IEEE 43rd Conference on Local Computer Networks Workshops (LCN Workshops) , pages=. 2018 , organization=

    work page 2018

  40. [40]

    Modeling, Control and Optimization of Water Systems: Systems Engineering Methods for Control and Decision Making Tasks , pages=

    Model based decision support systems , author=. Modeling, Control and Optimization of Water Systems: Systems Engineering Methods for Control and Decision Making Tasks , pages=. 2016 , publisher=

    work page 2016

  41. [41]

    Forecast informed reservoir operations using ensemble streamflow predictions for a multipurpose reservoir in

    Delaney, Chris J and Hartman, Robert K and Mendoza, John and Dettinger, Michael and Delle Monache, Luca and Jasperse, Jay and Ralph, F Martin and Talbot, Cary and Brown, James and Reynolds, David and others , journal=. Forecast informed reservoir operations using ensemble streamflow predictions for a multipurpose reservoir in. 2020 , publisher=

    work page 2020

  42. [42]

    Machine-learning algorithms for forecast-informed reservoir operation (

    Zarei, Manizhe and Bozorg-Haddad, Omid and Baghban, Sahar and Delpasand, Mohammad and Goharian, Erfan and Lo. Machine-learning algorithms for forecast-informed reservoir operation (. Scientific reports , volume=. 2021 , publisher=

    work page 2021

  43. [43]

    Leveraging open source software and parallel computing for model predictive control of urban drainage systems using

    Sadler, Jeffrey M and Goodall, Jonathan L and Behl, Madhur and Morsy, Mohamed M and Culver, Teresa B and Bowes, Benjamin D , journal=. Leveraging open source software and parallel computing for model predictive control of urban drainage systems using. 2019 , publisher=

    work page 2019

  44. [44]

    Journal of Water Resources Planning and Management , volume=

    Combining model predictive control with a reduced genetic algorithm for real-time flood control , author=. Journal of Water Resources Planning and Management , volume=. 2018 , publisher=

    work page 2018

  45. [45]

    Dimension reduction of decision variables for multireservoir operation:

    Chen, Duan and Leon, Arturo S and Gibson, Nathan L and Hosseini, Parnian , journal=. Dimension reduction of decision variables for multireservoir operation:. 2016 , publisher=

    work page 2016

  46. [46]

    Dynamic framework for intelligent control of river flooding:

    Leon, Arturo S and Kanashiro, Elizabeth A and Valverde, Rachelle and Sridhar, Venkataramana , journal=. Dynamic framework for intelligent control of river flooding:. 2014 , publisher=

    work page 2014

  47. [47]

    Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , pages=

    High-resolution image synthesis with latent diffusion models , author=. Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , pages=

    work page

  48. [48]

    International conference on machine learning , pages=

    Learning transferable visual models from natural language supervision , author=. International conference on machine learning , pages=. 2021 , organization=

    work page 2021

  49. [49]

    Hierarchical Text-Conditional Image Generation with CLIP Latents

    Hierarchical text-conditional image generation with clip latents , author=. arXiv preprint arXiv:2204.06125 , volume=

    work page internal anchor Pith review Pith/arXiv arXiv

  50. [50]

    Advances in neural information processing systems , volume=

    Photorealistic text-to-image diffusion models with deep language understanding , author=. Advances in neural information processing systems , volume=

    work page

  51. [51]

    Nature , volume=

    The quiet revolution of numerical weather prediction , author=. Nature , volume=. 2015 , publisher=

    work page 2015

  52. [52]

    Quarterly Journal of the Royal Meteorological Society , volume=

    Analysis methods for numerical weather prediction , author=. Quarterly Journal of the Royal Meteorological Society , volume=. 1986 , publisher=

    work page 1986

  53. [53]

    Visibility forecasting using autoregressive integrated moving average (

    Salman, Afan Galih and Kanigoro, Bayu , journal=. Visibility forecasting using autoregressive integrated moving average (. 2021 , publisher=

    work page 2021

  54. [54]

    Use of the autoregressive integrated moving average (

    Lai, Yuchuan and Dzombak, David A , journal=. Use of the autoregressive integrated moving average (

    work page

  55. [55]

    The effectiveness of extreme rainfall alerts in predicting surface water flooding in

    Hurford, AP and Priest, SJ and Parker, DJ and Lumbroso, DM , journal=. The effectiveness of extreme rainfall alerts in predicting surface water flooding in. 2012 , publisher=

    work page 2012

  56. [56]

    Fourcastnet:

    Pathak, Jaideep and Subramanian, Shashank and Harrington, Peter and Raja, Sanjeev and Chattopadhyay, Ashesh and Mardani, Morteza and Kurth, Thorsten and Hall, David and Li, Zongyi and Azizzadenesheli, Kamyar and others , journal=. Fourcastnet:

    work page

  57. [57]

    International Journal of Computer and Systems Engineering , volume=

    Time Series Forecasting Using Various Deep Learning Models , author=. International Journal of Computer and Systems Engineering , volume=

    work page

  58. [58]

    Deep learning models for water stage predictions in

    Shi, Jimeng and Yin, Zeda and Myana, Rukmangadh and Ishtiaq, Khandker and John, Anupama and Obeysekera, Jayantha and Leon, Arturo and Narasimhan, Giri , journal=. Deep learning models for water stage predictions in

    work page

  59. [59]

    arXiv preprint arXiv:2310.07631 , year=

    Graph Transformer Network for Flood Forecasting with Heterogeneous Covariates , author=. arXiv preprint arXiv:2310.07631 , year=

    work page arXiv

  60. [60]

    arXiv preprint arXiv:2310.19166 , year=

    The Power of Explainability in Forecast-Informed Deep Learning Models for Flood Mitigation , author=. arXiv preprint arXiv:2310.19166 , year=

    work page arXiv

  61. [61]

    Simulation of Flood Inundation Extent by Integration of

    Debbarma, Sagar and Dey, Subhajit and Bandyopadhyay, Arnab and Bhadra, Aditi , journal=. Simulation of Flood Inundation Extent by Integration of. 2024 , publisher=

    work page 2024

  62. [62]

    Advances in neural information processing systems , volume=

    Masked autoencoders as spatiotemporal learners , author=. Advances in neural information processing systems , volume=

    work page

  63. [63]

    Acute health risks to community hand-pumped groundwater supplies following

    Rivett, Michael O and Tremblay-Levesque, Laurent-Charles and Carter, Ruth and Thetard, Rudi CH and Tengatenga, Morris and Phoya, Ann and Mbalame, Emma and Mchilikizo, Edwin and Kumwenda, Steven and Mleta, Prince and others , journal=. Acute health risks to community hand-pumped groundwater supplies following. 2022 , publisher=

    work page 2022

  64. [64]

    Modeling unsteady and steady 1

    Gomes Jr, Marcus N and R. Modeling unsteady and steady 1. Environmental Modelling & Software , volume=. 2023 , publisher=

    work page 2023

  65. [65]

    Hydrology and Earth System Sciences Discussions , volume=

    Deep learning methods for flood mapping: a review of existing applications and future research directions , author=. Hydrology and Earth System Sciences Discussions , volume=. 2022 , publisher=

    work page 2022

  66. [66]

    Drivers of tidal flow variability in the

    Rahman, Motiur and Ali, Md Shahjahan , journal=. Drivers of tidal flow variability in the. 2024 , publisher=

    work page 2024

  67. [67]

    Water Resources Research , volume=

    Development of a fast and accurate hybrid model for floodplain inundation simulations , author=. Water Resources Research , volume=. 2023 , publisher=

    work page 2023

  68. [68]

    Journal of Hydroinformatics , volume=

    Urban flash flood forecast using support vector machine and numerical simulation , author=. Journal of Hydroinformatics , volume=. 2018 , publisher=

    work page 2018

  69. [69]

    Science of the Total Environment , volume=

    An ensemble prediction of flood susceptibility using multivariate discriminant analysis, classification and regression trees, and support vector machines , author=. Science of the Total Environment , volume=. 2019 , publisher=

    work page 2019

  70. [70]

    Water , volume=

    Regional inundation forecasting using machine learning techniques with the internet of things , author=. Water , volume=. 2020 , publisher=

    work page 2020

  71. [71]

    Reviews of Geophysics , volume=

    Recent advances and new frontiers in riverine and coastal flood modeling , author=. Reviews of Geophysics , volume=. 2023 , publisher=

    work page 2023

  72. [72]

    Environmental Modelling & Software , volume=

    A rapid flood inundation modelling framework using deep learning with spatial reduction and reconstruction , author=. Environmental Modelling & Software , volume=. 2021 , publisher=

    work page 2021

  73. [73]

    Flood inference simulation using surrogate modelling for the

    Castro-Gama, Mario E and Popescu, Ioana and Li, Shengyang and Mynett, Arthur and van Dam, Arthur , journal=. Flood inference simulation using surrogate modelling for the. 2014 , publisher=

    work page 2014

  74. [74]

    Integration of

    Peker,. Integration of. Sustainability , volume=. 2024 , publisher=

    work page 2024

  75. [75]

    Journal of Hydroinformatics , volume=

    Flood mitigation in coastal urban catchments using real-time stormwater infrastructure control and reinforcement learning , author=. Journal of Hydroinformatics , volume=. 2021 , publisher=

    work page 2021

  76. [76]

    The open real-time control (

    Schwanenberg, D and Becker, BPJ and Xu, M , journal=. The open real-time control (. 2015 , publisher=

    work page 2015

  77. [77]

    An overview of gradient descent optimization algorithms

    An overview of gradient descent optimization algorithms , author=. arXiv preprint arXiv:1609.04747 , year=

    work page internal anchor Pith review Pith/arXiv arXiv

  78. [78]

    South Florida Water Management District , howpublished =

    work page

  79. [79]

    Advances in neural information processing systems , volume=

    Attention is all you need , author=. Advances in neural information processing systems , volume=

    work page

  80. [80]

    Rolling window time series prediction using

    Li, Lei and Noorian, Farzad and Moss, Duncan JM and Leong, Philip HW , booktitle=. Rolling window time series prediction using. 2014 , organization=

    work page 2014

Showing first 80 references.